Chipmaker Xilinx Inc. said today it has teamed up with an artificial intelligence startup called Spline.ai to create a medical X-ray classification model that it says can improve prediction rates for COVID-19 and pneumonia using AI at the edge.

The model was built using more than 30,000 verified pneumonia and 500 COVID-19 X-ray images that were collected at public health and research facilities such as the National Institutes of Health, MIT and Stanford University, and trained using Amazon Web Services Inc.’s SageMaker AI service. The model is designed to aid respiratory disease detection, where studies show up to 30% of missed diagnoses are the result of human fatigue.

Xilinx is also making available a complete reference kit for medical firms to get started with the model, which includes a Xilinx UltraScale+ field-programmable gate array chip. FPGAs are computer chips that can be reprogrammed after they’ve been built, so they can be optimized for specific tasks on the fly.

In this case, Xilinx said the FPGA functions as a tensor accelerator that runs the AI model at the edge. The model itself can be deployed using the AWS Greengrass “internet of things” service. Xilinx said this configuration enables remote learning and geographically distributed inference, which means the image classification can be done onsite.

Instructions on the FPGA chip can be written using the Python programming language, so the model can be adapted to different application requirements.

Xilinx and Spline.ai’s reference kit is meant to help clinical researchers develop radiology workflows for diagnostic applications, the companies said.

“As the model can be easily adapted to similar clinical and diagnostic applications, medical equipment makers and health care providers are empowered to swiftly develop future clinical and radiological applications using the reference design kit,” said Kapil Shankar, vice president of marketing at Xilinx’s core markets group.

The ability of Xilinx’s FPGAs to power AI workloads is believed to be one of the primary reasons why Advanced Micro Devices Inc. is reportedly interested in acquiring the company. Last week, The Wall Street Journal said AMD was in talks to buy Xilinx in a deal worth about $30 billion in order to add FPGAs to its product portfolio and help it better compete with Intel Corp., which also makes FPGAs.

Image: PublicDomainPictures/Pixabay